interactive-serving
Operate TurboDiffusion's interactive T2V/I2V terminal server with safe launch validation, TUI commands, and runtime parameter handling.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- interactive-serving
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Operate TurboDiffusion's interactive T2V/I2V terminal server with safe launch validation, TUI commands, and runtime parameter handling.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/cli-reference.md5.4 KB
- 参考文档references/troubleshooting.md6.4 KB
- 参考文档references/tui-server.md5.5 KB
03 SKILL.md 正文
TurboDiffusion interactive serving
Use this sub-skill when the task is to launch or troubleshoot TurboDiffusion's interactive terminal server for repeated text-to-video (T2V) or image-to-video (I2V) generations while keeping loaded model objects alive across prompts.
Route here for
- Choosing an installed CLI vs Python-module TUI launch command.
- Validating T2V
--dit_pathvs I2V--high_noise_model_pathand--low_noise_model_pathbefore opening the TUI. - Explaining the TUI prompt flow, slash commands, output path prompts, and I2V image path prompts.
- Adjusting runtime parameters with
/setwithout restarting the server. - Diagnosing source-layout import errors, missing mode-specific model paths, invalid resolution/aspect ratio selections,
/setvalidation errors, and interactive cancellation.
Route elsewhere
- One-shot, non-interactive T2V/I2V scripts, prompt files, and direct
--save_pathgeneration: use the siblingvideo-inferencesub-skill. - CUDA extension builds,
turbo_diffusion_ops, SageSLA/SpargeAttn installation, quantized-linear backend failures, or FastNorm/custom-op checks: use the siblingacceleration-backendssub-skill. - Checkpoint conversion, training, model merging, or quantized checkpoint export: use the sibling
training-and-checkpointssub-skill.
Runtime files
- TUI server guide: launch flow, model residency behavior, slash commands, prompt/image/output handling, and runtime-adjustable parameters.
- CLI reference: public launch methods, argument tables, mode-specific requirements, defaults, and resolution/aspect-ratio validation.
- Troubleshooting: failure matrix for import quirks, missing model paths, invalid CLI values,
/seterrors, cancellation, and backend-routed problems. - Command renderer: safe dry-run helper that renders, but never executes, a T2V or I2V TUI launch command.
Quick operating pattern
- Decide whether the user needs the installed
turbodiffusion-serveentry point orpython -m turbodiffusion.serve. - Use the command renderer or the tables in CLI reference to build a mode-correct launch.
- For source-layout installs, prepend a public
PYTHONPATHentry that exposes TurboDiffusion's top-levelimaginaire,rcm, and inference helper modules; a common source checkout usesPYTHONPATH=turbodiffusion. - Once the TUI starts, keep it running for multiple prompts so VAE/model objects remain loaded instead of paying startup cost per generation.
- Use Troubleshooting before changing backend packages or rerouting to another sub-skill.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.